Microsoft 365 was down for two days over an authentication configuration

Microsoft 365 spent two days half-broken, and the cause was a login system that stopped recognizing its own services.

What failed

The outage began on August 31, 2026 and dragged into September 1. Exchange Online went first: mail failed to send or arrive, and sign-ins broke. The list then grew to Teams, OneDrive, SharePoint, Purview, Defender XDR and Microsoft 365 Copilot, whose prompts fail when they need a user’s mailbox as context. Microsoft tracked the incident as EX1464935 and then MO1465074 and described the cause as an “issue within a core authentication configuration used by multiple Microsoft 365 services.”

The recovery

Recovery was slow. Microsoft said mail flow and search came back first, followed by “a period of extended monitoring,” and by September 2 it reported service availability above 99% with a small set of users still affected. NHS mail support in the UK posted its own ongoing alert while the disruption lasted.

German IT blog BornCity and others reported that the trigger was a certificate that was not renewed on time. Microsoft has confirmed only the configuration issue, not the certificate.

Why one failure took everything

The practical lesson is the one every large outage teaches and every company forgets: when a single authentication layer serves email, chat, files and the AI assistant, they all fail together. Copilot, sold as the tool that knows your work, cannot answer when it cannot read your inbox. The dependency that makes the assistant useful is the same one that took it down.